Multi-Agent Apps

Welcome to Multi-Agent Apps – a collection of examples showcasing creative ways to use Microsoft Autogen for building AI-powered applications or performing complex AI-driven tasks. Whether you're a developer exploring multi-agent collaboration or looking for inspiration for your next AI project, this repository offers structured, reusable templates for multi-agent systems.

---

Features

• Multi-Agent Collaboration: Examples demonstrate agents specializing in tasks, collaborating effectively to achieve goals.
• Custom Termination Conditions: Define when tasks or projects are marked as complete using termination triggers.
• Dynamic Teams: Organize agents into flexible groups like round-robin chats or task-specific collaborations.
• Integration with Microsoft Autogen: Leveraging the powerful autogen framework to build modular, scalable AI systems.
• Interactive Console UI: Observe real-time collaboration and task updates through a simple UI.

---

Example: Building a Financial Budgeting Website

This example demonstrates how multiple agents collaborate to create a fully functional financial budgeting website with six pages:

• Index Page
• User Intake Questionnaire
• Income Tracking
• Monthly Budget
• Net Worth Calculator
• Financial Health Assessment

Code Snippet

from autogen_agentchat.agents import AssistantAgent
from autogen_agentchat.conditions import TextMentionTermination
from autogen_agentchat.teams import RoundRobinGroupChat
from autogen_agentchat.ui import Console
from autogen_ext.models.openai import OpenAIChatCompletionClient

# Define the Planner/Moderator Agent
planner_agent = AssistantAgent(
    "planner",
    model_client=OpenAIChatCompletionClient(model="gpt-4", api_key="your_api_key_here"),
    system_message="You are the Planner and Moderator...",
)

# Example: Index Agent
index_agent = AssistantAgent(
    "index_agent",
    model_client=model_client,
    system_message="You are responsible for creating the Index page..."
)

# Team and Termination Setup
termination_condition = TextMentionTermination("PROJECT COMPLETED")
team = RoundRobinGroupChat(
    [planner_agent, index_agent],
    termination_condition=termination_condition
)

# Async Function to Run the Task
async def main():
    await Console(team.run_stream(task="Develop the financial budgeting website..."))

if __name__ == "__main__":
    asyncio.run(main())